Characterizing time-of-day physical activity patterns across cognitive trajectories in older adults: a longitudinal cohort study

Declining cognitive function is a growing health challenge, yet how cognitive decline influences time-of-day physical activity (PA) patterns remains underexplored. Among cognitively unimpaired participants from the Atherosclerosis Risk in Communities Study, we modeled general and domain-specific cognitive trajectories and related them to accelerometry-derived PA a median of 8.2 years later in 811 participants (mean baseline age 73.1 years) using function-on-scalar and multivariable regression. Slower cognitive decline was associated with more favorable midday PA (10 AM–3 PM), including higher movement volume, less sedentary time, fewer interruptions to active periods, and greater activity variability. In adjusted linear regression models, associations with PA patterns derived from a window as short as 11 AM-1 PM were comparable to those with 24-hour measures. The executive function trajectory exhibited the strongest associations with PA patterns, and over the longest period of the day; the language trajectory showed non-linear associations, and the peak association for the memory trajectory with PA occurred in the late morning (10:30–11:30 AM). Associations between cognitive decline and activity variability were stronger than for other PA metrics. Our findings highlight the potential of PA patterns, particularly activity variability, as a non-invasive way to help monitor cognitive decline before clinical impairment emerges.